Yes, sure - I'll link the source archive when I get home. It's quite a simple program so modifying/adding things shouldn't be an issue.
For actually investigating failed cases though it might be better to have some error magnitude measurement which makes sense for each test. I didn't bother with this since it was made just to detect faulty chips (IEEE-754 formatted results take part in a running hash), and error magnitude probably won't make sense for every kind of test (such as fmod) - but a subset extra measurements for this might be helpful...
D.